CN112706655B - Energy balance control method and system for new energy vehicles based on intelligent Internet of Things - Google Patents
Energy balance control method and system for new energy vehicles based on intelligent Internet of Things Download PDFInfo
- Publication number
- CN112706655B CN112706655B CN202110090419.0A CN202110090419A CN112706655B CN 112706655 B CN112706655 B CN 112706655B CN 202110090419 A CN202110090419 A CN 202110090419A CN 112706655 B CN112706655 B CN 112706655B
- Authority
- CN
- China
- Prior art keywords
- electric quantity
- sequence
- access time
- soh
- soh estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000011159 matrix material Substances 0.000 claims abstract description 69
- 238000011217 control strategy Methods 0.000 claims abstract description 5
- 239000013598 vector Substances 0.000 claims description 84
- 238000004458 analytical method Methods 0.000 claims description 47
- 230000008859 change Effects 0.000 claims description 25
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012986 modification Methods 0.000 claims description 2
- 230000004048 modification Effects 0.000 claims description 2
- 230000005611 electricity Effects 0.000 claims 2
- 230000003190 augmentative effect Effects 0.000 claims 1
- 230000036541 health Effects 0.000 abstract description 3
- 230000032683 aging Effects 0.000 description 9
- 238000012549 training Methods 0.000 description 9
- 238000007599 discharging Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000010606 normalization Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
- B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
- B60L58/18—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries of two or more battery modules
- B60L58/22—Balancing the charge of battery modules
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
- B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
- B60L58/16—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries responding to battery ageing, e.g. to the number of charging cycles or the state of health [SoH]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
- H02J7/0013—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries acting upon several batteries simultaneously or sequentially
- H02J7/0014—Circuits for equalisation of charge between batteries
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Secondary Cells (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
本发明公开了一种基于智能物联网的新能源汽车能源均衡控制方法及系统。方法包括：利用第一SOH估计网络分别对新能源汽车各电池组进行SOH估计得到第一SOH估计序列；根据第一SOH估计序列、汽车启动功率，生成第一接入时间矩阵；根据接入时间窗口尺寸、电池组接入约束对第一接入时间矩阵进行扩充修正，得到第二接入时间张量；根据第一SOH估计序列、第二接入时间张量，利用第二SOH估计网络对各电池组进行SOH估计得到第二SOH估计序列；根据各电池组的第二SOH估计序列，计算新能源汽车的能源均衡度，根据最高能源均衡度对应的第二接入时间张量通道矩阵，得到新能源汽车电池组的均衡控制策略。利用本发明实现了汽车电池组能源的健康状态平衡。
The invention discloses a new energy vehicle energy balance control method and system based on the intelligent Internet of things. The method includes: using a first SOH estimation network to respectively perform SOH estimation on each battery group of a new energy vehicle to obtain a first SOH estimation sequence; generating a first access time matrix according to the first SOH estimation sequence and the starting power of the vehicle; and according to the access time The window size and the battery pack access constraints are expanded and corrected for the first access time matrix to obtain the second access time tensor; according to the first SOH estimation sequence and the second access time tensor, the second SOH estimation network is used to estimate the network pair. Each battery pack performs SOH estimation to obtain a second SOH estimation sequence; according to the second SOH estimation sequence of each battery pack, the energy balance degree of the new energy vehicle is calculated, and according to the second access time tensor channel matrix corresponding to the highest energy balance degree, The equilibrium control strategy of the battery pack of the new energy vehicle is obtained. The invention realizes the health state balance of the energy of the vehicle battery pack.
Description
Technical Field
The invention relates to the technical field of new energy, in particular to a new energy automobile energy balance control method and system based on an intelligent Internet of things.
Background
The active balancing is the mainstream direction of new energy battery energy balancing, and active balancing methods in the prior art include temperature balancing, charge and discharge balancing, load changing to increase uninterrupted output energy and the like. The existing active equalization method of the parallel battery pack has less research, the terminal voltage of the parallel battery pack is equalized mainly through measures such as discharging, and the like.
Disclosure of Invention
The invention aims to provide a new energy automobile energy balance control method and system based on an intelligent Internet of things, aiming at the defects in the prior art.
A new energy automobile energy balance control method based on an intelligent Internet of things comprises the following steps:
step 1, based on a charging curve of each battery pack of the new energy automobile, SOH estimation is respectively carried out on each battery pack by utilizing a first SOH estimation network to obtain a first SOH estimation sequence;
step 2, generating a first access time matrix according to the first SOH estimation sequence and the automobile starting power; expanding and correcting the first access time matrix according to the size of the access time window and the access constraint of the battery pack to obtain a second access time tensor;
step 3, according to the first SOH estimation sequence and the second access time tensor, carrying out SOH estimation on each battery pack by using a second SOH estimation network to obtain a second SOH estimation sequence corresponding to each channel of the second access time tensor;
and 4, calculating the energy balance degree of the new energy automobile according to the second SOH estimation sequence corresponding to each channel, and obtaining a balance control strategy of the new energy automobile battery pack according to the second access time tensor channel corresponding to the highest energy balance degree.
Further, the generating a first access time matrix according to the first SOH estimation sequence and the vehicle starting power includes:
selecting a plurality of battery packs with the largest SOH from the first SOH estimation sequence, wherein the selected battery packs can meet the requirement of automobile starting power and have the smallest number;
according to the number N of the selected battery packs1Generating a first access time matrix, wherein the size of the first access time matrix is K x M, K is the number of battery packs, M is the size of an access time window, and the first N of the first column in the first access time matrix1Each element is a first setting value.
Further, the expanding and modifying the first access time matrix according to the size of the access time window and the battery pack access constraint to obtain a second access time tensor comprises:
the battery pack access constraints include:
k is the number of battery packs, M is the access time window size, ea,k,mC is a first set value, and is an element of a second access time matrix tensor a channel k row and m column; the column of the first setting value element of the adjacent row of the same channel satisfies: the column of the next row of first setting value elements is not less than the column of the previous row of first setting value elements;
and expanding and correcting the first access time matrix according to the constraint of the battery pack to obtain a second access time tensor, wherein the size of the second access time tensor is A x K x M, and A is the number of channels.
Further, the second SOH estimation network comprises:
the initial SOH analysis network branch is used for analyzing the first SOH estimation sequence to obtain an initial SOH analysis vector;
the access time analysis network branch is used for analyzing the channel matrix of the second access time tensor to obtain an access time analysis vector;
the electric quantity change analysis network branch is used for analyzing an electric quantity change sequence corresponding to the channel matrix of the second access time tensor to obtain an electric quantity change analysis vector; the electric quantity change sequence is a difference value sequence of the initial electric quantity sequence and the final electric quantity sequence;
and the SOH estimation network branch is used for analyzing the characteristic vector after the initial SOH analysis vector, the access time analysis vector and the electric quantity change analysis vector are fused to obtain a second SOH estimation sequence.
Further, the final power sequence is obtained by:
step 3a, performing descending ordering on the electric quantity sequence corresponding to the first SOH estimation sequence to obtain an initial electric quantity sequence, and taking the obtained initial electric quantity sequence as an electric quantity sequence to be analyzed;
step 3b, taking a channel from the second access time tensor;
step 3c, obtaining an input electric quantity matrix, the number of power supply battery packs and power supply duration according to the obtained channel, the electric quantity sequence to be analyzed and the cycle times;
step 3d, analyzing the input electric quantity matrix, the number of power supply battery packs and the power supply duration by using the electric quantity estimation neural network to obtain an estimated electric quantity sequence, generating an electric quantity sequence to be analyzed according to the estimated electric quantity sequence, and turning to the step 3 c;
step 3e, circularly executing the steps 3c-3d until the number of the power supply battery packs is equal to that of the battery packs held by the automobile, and obtaining a final electric quantity sequence;
and 3f, circularly executing the steps 3b-3e until all channels are traversed to obtain a final electric quantity sequence set.
Further, the obtaining of the input electric quantity matrix according to the obtained channel and the electric quantity sequence to be analyzed includes:
and sequentially taking d electric quantity values from the electric quantity sequence to be analyzed to sequentially replace corresponding elements of the first set values in the front d rows of the taken channel, and returning other first set values in the taken channel to zero to obtain an input electric quantity matrix, wherein d is the number of the power supply battery packs determined according to the cycle number.
Further, the electric quantity estimation neural network includes:
the first network branch is used for analyzing the input electric quantity matrix to obtain a first characteristic vector;
the second network branch is used for analyzing the number of the power supply battery packs to obtain a second feature vector;
the third network branch is used for analyzing the power supply duration to obtain a third feature vector;
and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector and the third feature vector are fused to obtain an estimated electric quantity sequence.
Furthermore, the electric quantity estimation neural network further comprises a fourth network branch for analyzing the power supply power to obtain a fourth eigenvector; and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector, the third feature vector and the fourth feature vector are fused to obtain an estimated electric quantity sequence.
Further, the second SOH estimation network further includes a charging number analysis network branch for analyzing the charging number sequence to obtain a charging number analysis vector; and the SOH estimation branch is used for analyzing the characteristic vector after the initial SOH analysis vector, the electric quantity change analysis vector, the access time analysis vector and the charging frequency analysis vector are fused to obtain a second SOH estimation sequence.
A new energy automobile energy balance system based on an intelligent Internet of things comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein when the computer program is executed by the processor, the steps of any one of the methods are realized.
Compared with the prior art, the invention has the following beneficial effects:
the optimal access mode is selected by analyzing the health states of the parallel battery packs in different access modes, so that the energy balance of the new energy automobile is realized. The first SOH estimation network, the second SOH estimation network and the electric quantity estimation network provide basis for the equalization strategy, and the obtained equalization strategy can effectively equalize the aging degree of the battery pack. The spatial domain characteristics of the battery accessed with different electric quantities are reflected by combining an access time tensor and an input electric quantity matrix, and the precision of the electric quantity estimation neural network is improved; the precision of the electric quantity estimation neural network is further improved by combining the charging power; and the estimation precision of the second SOH estimation network is improved by combining the charging times.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a new energy automobile energy balance control method and system based on an intelligent Internet of things. FIG. 1 is a flow chart of the method of the present invention. The following description will be made by way of specific examples.
Example 1:
a new energy automobile energy balance control method based on an intelligent Internet of things comprises the following steps:
step 1, based on the charging curves of the battery packs of the new energy automobile, SOH estimation is respectively carried out on the battery packs by utilizing a first SOH estimation network to obtain a first SOH estimation sequence.
In the field of battery equalization, there have been many efforts on the equalization of power in the case of series-connected power supply units, but there have been few studies on the equalization of power between parallel-connected power supply units. The existing power equalization is mainly realized through passive equalization or active equalization. Passive equalization, which can realize electric quantity equalization but wastes part of electric energy, thus being not beneficial to saving energy; the active equalization realizes the power equalization by charging the power supply unit with low power or low voltage, and has the defect that although the power equalization is realized, the charging and discharging times of the power supply unit with low power or low voltage are increased, which is not beneficial to the equalization of the aging degree of the power supply unit. According to the invention, the access time of the power supply unit is controlled, the service time and the charging and discharging times of the power supply unit with large aging degree are reduced, and thus the balance of the aging degree is realized. The battery pack of the invention passes through the switch control module, and each battery pack can realize the control of access power supply through the switch control.
The charging curve is a change curve of terminal voltage of each battery pack or each single battery along with time when the new energy automobile battery pack is charged, reflects the relation among charging current, charging voltage and time, and is obtained through real-time monitoring. Since the charging voltage curves of the batteries with different aging degrees are different, the charging curves can represent the aging state of the batteries.
SOH, i.e., battery state of health, characterizes battery capacity fade,
in the formula CagedIs the current maximum capacity, C, of the batteryratedFor the rated capacity of the battery, the SOH estimation method comprises a direct discharge method, an internal resistance method, an electrochemical impedance analysis method, a model method and a voltage curve model method. According to the invention, SOH estimation is carried out according to the voltage curve, and the SOH state corresponding to the battery is obtained according to the charging voltage normalization curve corresponding to the battery. Because the battery pack at least comprises one single battery, one embodiment is to collect the charging curve of each single battery and obtain the average SOH state according to the SOH state of each single battery as the SOH state of the battery pack; in one embodiment, the charging curve of the battery pack is monitored by taking the battery pack as a unit, and the SOH state corresponding to the battery pack is obtained according to the charging voltage normalization curve corresponding to the battery pack. The first embodiment is more accurate and the second embodiment is more efficient.
The charging voltage normalization curve can be measured in an experimental environment and can also be realized through an intelligent Internet of things technology. The charging voltage normalization curve can be obtained based on the intelligent Internet of things, the charging curves of battery packs of the same model in a plurality of automobiles of the same model are obtained through the intelligent Internet of things, and the charging voltage normalization curve corresponding to the battery packs of the model is obtained according to the classification of the charging and discharging cycle times of the battery packs.
The network architecture of the first SOH estimation network is an encoder-full connectivity layer architecture. The input of the first SOH estimation network is a charging curve, and the output is a predicted SOH value, namely a first SOH; the specific details of the network training are as follows: the method comprises the steps of collecting charging curves of various batteries or battery packs with different aging degrees (or charging cycle times) as a training data set, using SOH corresponding to the battery packs as marking data, using a mean square error loss function as a loss function, and training by using the training data set until network convergence. When in use, according to the single battery in the battery packThe charging curve or the charging curve of the battery pack is input into a first SOH estimation network, and a first SOH estimation result corresponding to the battery pack can be obtained. The first SOH estimation results of the respective battery packs constitute a first SOH estimation sequence. Specifically, the first SOH estimation results are sorted in descending order according to size to form a first SOH sequence [ H [ ]1,H2,…,HK]A smaller sequence number indicates a larger SOH, i.e., a smaller degree of aging.
Step 2, generating a first access time matrix according to the first SOH estimation sequence and the automobile starting power; and expanding and correcting the first access time matrix according to the size of the access time window and the access constraint of the battery pack to obtain a second access time tensor.
Generating a first access time matrix according to the first SOH estimation sequence and the automobile starting power comprises the following steps: selecting a plurality of battery packs with the largest SOH from the first SOH estimation sequence, wherein the selected battery packs can meet the requirement of automobile starting power and have the smallest number; according to the number N of the selected battery packs1Generating a first access time matrix, wherein the size of the first access time matrix is K x M, K is the number of battery packs, M is the size of an access time window, and the first N of the first column in the first access time matrix1Each element is a first setting value. Specifically, the automobile starting power is the rated automobile starting power, and the automobile starting power is set to be P according to the acquisition of the controller of the new energy automobile and the factory parameters of the motors. P is the starting power of the automobile based on the first SOH sequencesThe number N of battery packs initially supplied is determined by the implementer1Generating a first access time matrix having a size K x M, wherein,
in the formula, T is the access time window duration, Δ T is the unit power supply period, and M is the access time window size, which represents that the battery pack must be completely accessed in M unit power supply periods, and the battery pack adding time can be determined according to T, Δ T and the battery pack access period value. The implementer may specifically set T, Δ T according to the implementation scenario, and preferably, in the case of 5 battery packs, T is set to 30min and Δ T is set to 3And (5) min. According to the number N of the battery packs for power supply1The first access time matrix is assigned a value,
the values of the elements of the first access time matrix are all taken as the first set value, preferably, the first set value is 1, and the values of other elements in the matrix are all 0.
The battery pack access constraints include:
k is the number of battery packs, M is the access time window size, ea,k,mC is an element of a second access time tensor a channel k row and m column, and is a first set value; the column of the first setting value element of the adjacent row of the same channel satisfies: the column in which the first setting value element of the next row is located is not smaller than the column in which the first setting value element of the previous row is located (in this embodiment, for the element with the value of 1 in the adjacent row, the column satisfies mk≤mk+1，mk、mk+1Respectively the column of the k row element 1 and the column of the k +1 row element 1); and expanding and correcting the first access time matrix according to the access constraint of the battery pack to obtain a second access time tensor, wherein the size of the second access time tensor is A x K x M, and A is the number of channels. The other elements of the second access time tensor are a second set value of 0. The above battery access constraints are for the access time tensor. It should be noted that, in the present invention, the battery packs are sequentially connected from large to small according to the SOH, that is, the battery pack newly connected to supply power is the battery pack with the largest SOH selected from the battery packs that are not connected to supply power each time.
And 3, according to the first SOH estimation sequence and the second access time tensor, carrying out SOH estimation on each battery pack by using a second SOH estimation network to obtain a second SOH estimation sequence corresponding to each channel of the second access time tensor.
The second SOH estimation network comprises: the initial SOH analysis network branch is used for analyzing the first SOH estimation sequence to obtain an initial SOH analysis vector; the access time analysis network branch is used for analyzing the channel matrix of the second access time tensor to obtain an access time analysis vector; the electric quantity change analysis network branch is used for analyzing an electric quantity change sequence corresponding to the channel matrix of the second access time tensor to obtain an electric quantity change analysis vector; the electric quantity change sequence is a difference value sequence of the initial electric quantity sequence and the final electric quantity sequence; and the SOH estimation network branch is used for analyzing the characteristic vector after the initial SOH analysis vector, the access time analysis vector and the electric quantity change analysis vector are fused to obtain a second SOH estimation sequence.
The input of the second SOH estimation network comprises a first SOH sequence, a second access time tensor and an electric quantity change sequence, and the output of the second SOH estimation network is a second SOH sequence; the structure of each network branch is that the first SOH sequence and the electric quantity change sequence correspond to respective full-connection networks, the channel matrix of the second access time tensor is connected with the encoder-full-connection network, and the eigenvectors output by each full-connection network are converted (combined) and then sent to the subsequent full-connection network; the second SOH estimation network training method comprises the following steps: the training set adopts a plurality of groups of channel matrixes of the first SOH sequence, the electric quantity change sequence and the second access time tensor, takes the SOH sequence obtained after recharging as label data, namely takes the SOH state of the battery pack after discharging is finished by the current SOH state of the battery pack according to the access time control method represented by the channel matrix of the second access time tensor as the label data, and specifically obtains the SOH state according to a charging curve obtained after discharging is finished and recharging is performed, and the loss function adopts an L2 loss function.
The electric quantity change sequence is obtained according to the initial electric quantity sequence and the final electric quantity sequence, and specifically, the final electric quantity sequence is subtracted from the initial electric quantity sequence. The initial electric quantity sequence can be obtained according to the first SOH estimation sequence, and the final electric quantity sequence is obtained in the following mode:
and 3a, sequencing the electric quantity sequence corresponding to the first SOH estimation sequence in a descending order to obtain an initial electric quantity sequence, and taking the obtained initial electric quantity sequence as an electric quantity sequence to be analyzed. According to a first SOH sequence [ H ]1,H2,…,HK]Obtaining an initial power sequence
The invention utilizesThe electric quantity estimation neural network analyzes the current electric quantity of a plurality of power supply battery packs accessed for power supply, and estimates the residual electric quantity of each battery pack after the plurality of power supply battery packs pass through corresponding power supply time. In the invention, all the battery packs are added in M unit power supply periods for power supply, so the electric quantity estimation neural network needs to be cycled for a plurality of times to obtain a final electric quantity sequence, and the cycle times are the adding batch times of the battery packs. For example, if the battery pack is supplied with power in 3 batches, the number of cycles is three. And in the first circulation, taking the initial electric quantity sequence as the electric quantity sequence to be analyzed.
And 3b, taking a channel from the second access time tensor. And selecting a matrix of one channel from the A channels of the second access time tensor, and recording the matrix as a second access time matrix.
And 3c, obtaining an input electric quantity matrix, the number of power supply battery packs and the power supply duration according to the acquired channel, the electric quantity sequence to be analyzed and the cycle number. Obtaining an input electric quantity matrix according to the obtained channel and the electric quantity sequence to be analyzed comprises the following steps: and sequentially taking d electric quantity values from the electric quantity sequence to be analyzed to sequentially replace corresponding elements of first set values in the front d rows of the taken channel, and returning other first set values in the taken channel to zero to obtain an input electric quantity matrix, wherein d is the number of the power supply battery packs determined according to the cycle number.
Specifically, an input electric quantity matrix is generated by combining the initial electric quantity sequence. The number N of the first column element value of the second access time matrix is 11Selecting the first N in the initial electric quantity sequence1An electric quantity value, a
And assigning the element with the value of 1 to the kth row element in the 1 st column in the second access time matrix according to the subscript k, and zeroing other elements with the value of 1. In subsequent cycles, the number of the battery packs accessed to supply power at each time can be obtained according to the cycle times, and the electric quantity of the battery pack of the electric quantity sequence to be analyzed is assigned to the corresponding element in the second access time matrix.
Acquiring a power supply battery pack in an initial time period according to the selected second access time matrixNumber N1And initial period length Δ t1From m, in particular1Starting from the column, the second access time matrix is analyzed column by column, the next column containing element 1 is recorded as mth column2Column, then (m)2-m1)*ΔT＝Δt1In the first cycle m11. By analogy, in the subsequent cycle, the power supply duration of the battery pack which is currently accessed to supply power can be obtained according to the unit power supply period number of the interval.
And 3d, analyzing the input electric quantity matrix, the number of power supply battery packs and the power supply duration by using the electric quantity estimation neural network to obtain an estimated electric quantity sequence, generating an electric quantity sequence to be analyzed according to the estimated electric quantity sequence, and turning to the step 3 c. Specifically, an electric quantity sequence to be analyzed is generated according to the estimated electric quantity sequence, that is, an initial electric quantity value of a battery pack to be connected with power supply in the next cycle is given to a corresponding element of the estimated electric quantity sequence, so that the electric quantity sequence to be analyzed is obtained. The battery packs selected to be connected with the power supply in each batch are the largest SOH battery packs which are not connected with the power supply at present.
The electric quantity estimation neural network comprises: the first network branch is used for analyzing the input electric quantity matrix to obtain a first characteristic vector; the second network branch is used for analyzing the number of the power supply battery packs to obtain a second feature vector; the third network branch is used for analyzing the power supply duration to obtain a third feature vector; and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector and the third feature vector are fused to obtain an estimated electric quantity sequence.
The first network branch comprises an encoder and a full-connection network, the second network branch comprises a full-connection network, the power supply time comprises the full-connection network, the three network branches sequentially output a first feature vector, a second feature vector and a third feature vector, and the three feature vectors are fused to obtain an electric quantity prediction feature vector. And the electric quantity prediction characteristic vector is sent to an electric quantity prediction network, and an estimated electric quantity sequence is output. And a plurality of values arranged in the front in the estimated electric quantity sequence after each circulation are not zero, wherein the number of the values is the number of the battery packs accessing to the power supply under the current circulation. After the first cycle, the power is estimatedOnly the first N in the quantitative sequence1Each element is not 0.
The training method of the electric quantity estimation network comprises the following steps: the training set comprises a plurality of groups of input electric quantity matrixes, the number of power supply units and power supply duration corresponding to different aging degrees and different access time matrixes, and the labeled data is an electric quantity sequence obtained by monitoring after the power supply duration; the loss function is a mean square error loss function.
In addition, in order to further improve the accuracy of the electric quantity estimation neural network, a network branch can be arranged to analyze the first SOH sequence, the characteristic vectors obtained by the network branch are fused with the first, second and third characteristic vectors, and then the electric quantity is sent to the electric quantity prediction network to be analyzed to obtain the estimated electric quantity sequence. This is done to improve the accuracy of estimating the power sequence by taking into account the influence of factors such as the internal resistance of the battery. Although the influence of the internal resistance of the battery is considered in the second SOH estimation network, the influence is added into the electric quantity estimation neural network, and the training efficiency and the network output accuracy of the second SOH estimation network are improved.
And 3e, circularly executing the steps 3c-3d until the number of the power supply battery packs is equal to that of the battery packs held by the automobile, and obtaining a final electric quantity sequence.
Specifically, the first loop has been described in detail above, and the second loop is taken as an example for the loop execution of step 3e, so as to make the implementation process of the present embodiment more clear to the implementer. In the second circulation, the number N of the power supply battery packs in the second time period is obtained according to the step 3c mode2And corresponding power supply duration deltat2And generating an input electric quantity matrix required by the secondary cycle according to the selected access time matrix in the step 3c mode, wherein if no element of a certain row has an element value of 1, the row is not assigned, and the step 3d is executed to output an estimated electric quantity sequence corresponding to the secondary cycle. And performing 3c-3d in a subsequent cycle until the number of the battery packs accessed to supply power is K, and obtaining a corresponding final electric quantity sequence.
And 3f, circularly executing the steps 3b-3e until all channels are traversed to obtain a final electric quantity sequence set. And analyzing each channel in the second access time tensor to obtain a final electric quantity sequence set.
The final electric quantity sequence set obtained by the battery pack in the current SOH state according to different access control modes is obtained in the steps 3a-3 f. And according to the final electric quantity sequence set, obtaining a corresponding electric quantity change sequence set, and sequentially inputting the first SOH estimation sequence, the second access time matrix and the corresponding electric quantity change sequence into a second SOH estimation network for analysis to obtain second SOH sequences under different access control modes.
And 4, calculating the energy balance degree of the new energy automobile according to the second SOH estimation sequence corresponding to each channel, and obtaining a balance control strategy of the new energy automobile battery pack according to the second access time tensor channel corresponding to the highest energy balance degree.
And calculating the energy balance degree of the new energy automobile according to the second SOH sequence. The balance degree reflects the dispersion degree of data in the second SOH sequence, preferably, the dispersion degree of the sequence is reflected by variance or standard deviation, and the second access time matrix corresponding to the second SOH sequence with the minimum variance is selected, so that the balance control strategy of the new energy automobile battery pack can be obtained.
Example 2:
in order to improve the precision of the electric quantity estimation neural network, the electric quantity estimation neural network further comprises a fourth network branch for analyzing the power supply power to obtain a fourth feature vector; and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector, the third feature vector and the fourth feature vector are fused to obtain an estimated electric quantity sequence.
Example 3:
the embodiment provides a new energy automobile energy balance control method based on an intelligent internet of things, which is improved based on embodiment 1, and the second SOH estimation network further comprises a charging time analysis network branch for analyzing the charging time sequence to obtain a charging time analysis vector. The charging times of each battery pack are the charging times of each battery pack receiving BMS electric quantity equalizing module in the access mode corresponding to the second access time matrix,obtained by monitoring, the charging times of each battery form a charging time sequence [ B1,B2,…,BK]. And the SOH estimation network branch in the second SOH estimation network is used for analyzing the characteristic vector after the initial SOH analysis vector, the electric quantity change analysis vector, the access time analysis vector and the charging frequency analysis vector are fused to obtain a second SOH estimation sequence.
Example 4:
the embodiment provides a new energy automobile energy balancing system based on an intelligent internet of things, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the steps of any one of the methods are realized.
The above embodiments are merely preferred embodiments of the present invention, which should not be construed as limiting the present invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. The new energy automobile energy balance control method based on the intelligent Internet of things is characterized by comprising the following steps:
step 1, based on a charging curve of each battery pack of the new energy automobile, SOH estimation is respectively carried out on each battery pack by utilizing a first SOH estimation network to obtain a first SOH estimation sequence;
step 2, generating a first access time matrix according to the first SOH estimation sequence and the automobile starting power; expanding and correcting the first access time matrix according to the size of the access time window and the access constraint of the battery pack to obtain a second access time tensor;
step 3, according to the first SOH estimation sequence and the second access time tensor, carrying out SOH estimation on each battery pack by using a second SOH estimation network to obtain a second SOH estimation sequence corresponding to each channel of the second access time tensor;
and 4, calculating the energy balance degree of the new energy automobile according to the second SOH estimation sequence corresponding to each channel, and obtaining a balance control strategy of the new energy automobile battery pack according to the second access time tensor channel corresponding to the highest energy balance degree.
2. The method of claim 1, wherein generating a first access time matrix based on the first sequence of SOH estimates, vehicle starting power, comprises:
selecting a plurality of battery packs with the largest SOH from the first SOH estimation sequence, wherein the selected battery packs can meet the requirement of automobile starting power and have the smallest number;
according to the selected number of the battery packs
Generating a first access time matrix, wherein the size of the first access time matrix is K x M, K is the number of battery packs, M is the size of an access time window, and the front of a first column in the first access time matrix
Each element is a first setting value.
3. The method of claim 1, wherein the augmented modification of the first access time matrix based on access time window size and battery access constraints to obtain the second access time tensor comprises:
the battery pack access constraints include:
,
k is the number of the battery packs, M is the size of the access time window,
for the second access time matrix tensor
Elements of channel k rows and m columns, c beingA first set value; the column of the first setting value element of the adjacent row of the same channel satisfies: the column of the next row of first setting value elements is not less than the column of the previous row of first setting value elements;
the first access time matrix is expanded and corrected according to the constraint of the battery pack to obtain a second access time tensor, and the size of the second access time tensor is
，
Is the number of channels.
4. The method of claim 1, wherein the second SOH estimation network comprises:
the initial SOH analysis network branch is used for analyzing the first SOH estimation sequence to obtain an initial SOH analysis vector;
the access time analysis network branch is used for analyzing the channel matrix of the second access time tensor to obtain an access time analysis vector;
the electric quantity change analysis network branch is used for analyzing an electric quantity change sequence corresponding to the channel matrix of the second access time tensor to obtain an electric quantity change analysis vector; the electric quantity change sequence is a difference value sequence of the initial electric quantity sequence and the final electric quantity sequence;
and the SOH estimation network branch is used for analyzing the characteristic vector after the initial SOH analysis vector, the access time analysis vector and the electric quantity change analysis vector are fused to obtain a second SOH estimation sequence.
5. The method according to claim 4, characterized in that the final sequence of electrical quantities is obtained by:
step 3a, performing descending ordering on the electric quantity sequence corresponding to the first SOH estimation sequence to obtain an initial electric quantity sequence, and taking the obtained initial electric quantity sequence as an electric quantity sequence to be analyzed;
step 3b, taking a channel from the second access time tensor;
step 3c, obtaining an input electric quantity matrix, the number of power supply battery packs and power supply duration according to the obtained channel, the electric quantity sequence to be analyzed and the cycle times;
step 3d, analyzing the input electric quantity matrix, the number of power supply battery packs and the power supply duration by using the electric quantity estimation neural network to obtain an estimated electric quantity sequence, generating an electric quantity sequence to be analyzed according to the estimated electric quantity sequence, and turning to the step 3 c;
step 3e, circularly executing the steps 3c-3d until the number of the power supply battery packs is equal to that of the battery packs held by the automobile, and obtaining a final electric quantity sequence;
and 3f, circularly executing the steps 3b-3e until all channels are traversed to obtain a final electric quantity sequence set.
6. The method of claim 5, wherein the obtaining of the input electricity quantity matrix according to the taken channel and the electricity quantity sequence to be analyzed comprises:
and sequentially taking d electric quantity values from the electric quantity sequence to be analyzed to sequentially replace corresponding elements of the first set values in the front d rows of the taken channel, and returning other first set values in the taken channel to zero to obtain an input electric quantity matrix, wherein d is the number of the power supply battery packs determined according to the cycle number.
7. The method of any one of claims 5-6, wherein the charge estimation neural network comprises:
the first network branch is used for analyzing the input electric quantity matrix to obtain a first characteristic vector;
the second network branch is used for analyzing the number of the power supply battery packs to obtain a second feature vector;
the third network branch is used for analyzing the power supply duration to obtain a third feature vector;
and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector and the third feature vector are fused to obtain an estimated electric quantity sequence.
8. The method of claim 7, wherein the power estimation neural network further comprises a fourth network branch for analyzing the supply power to obtain a fourth eigenvector; and the electric quantity prediction network is used for analyzing the feature vector after the first feature vector, the second feature vector, the third feature vector and the fourth feature vector are fused to obtain an estimated electric quantity sequence.
9. The method of claim 4, wherein said second SOH estimation network further comprises a charging times analysis network branch for analyzing a charging times sequence to obtain a charging times analysis vector; and the SOH estimation network branch is used for analyzing the characteristic vector after the initial SOH analysis vector, the electric quantity change analysis vector, the access time analysis vector and the charging frequency analysis vector are fused to obtain a second SOH estimation sequence.
10. An intelligent internet of things-based new energy automobile energy balance system, comprising a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of the method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110090419.0A CN112706655B (en) | 2021-01-22 | 2021-01-22 | Energy balance control method and system for new energy vehicles based on intelligent Internet of Things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110090419.0A CN112706655B (en) | 2021-01-22 | 2021-01-22 | Energy balance control method and system for new energy vehicles based on intelligent Internet of Things |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112706655A CN112706655A (en) | 2021-04-27 |
| CN112706655B true CN112706655B (en) | 2021-09-07 |
Family
ID=75549403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110090419.0A Active CN112706655B (en) | 2021-01-22 | 2021-01-22 | Energy balance control method and system for new energy vehicles based on intelligent Internet of Things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112706655B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114103707B (en) * | 2021-12-06 | 2024-01-26 | 黄淮学院 | Intelligent energy control method and system based on artificial intelligence and Internet of things |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2929410B1 (en) * | 2008-03-28 | 2010-04-09 | Inst Francais Du Petrole | METHOD FOR ESTIMATING NON-MEASURABLE CHARACTERISTICS OF AN ELECTROCHEMICAL SYSTEM |
| KR101223735B1 (en) * | 2011-04-07 | 2013-01-21 | 로베르트 보쉬 게엠베하 | Battery management system and control method thereof |
| DE102013209404A1 (en) * | 2013-05-22 | 2014-12-11 | Robert Bosch Gmbh | Cell arrangement with a plurality of electrochemical cells and method for operating the same |
| CN106026226B (en) * | 2016-05-19 | 2019-07-09 | 东软集团股份有限公司 | Battery equalization method, device and battery system |
| CN111193306B (en) * | 2020-02-20 | 2021-04-13 | 山东大学 | Battery health state balancing method and system of modular energy storage battery grid-connected system |
-
2021
- 2021-01-22 CN CN202110090419.0A patent/CN112706655B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112706655A (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109256834B (en) | Active balancing method of battery pack based on battery state of health and state of charge | |
| Caspar et al. | Systematic comparison of active balancing: A model-based quantitative analysis | |
| Abdollahi et al. | Optimal charging for general equivalent electrical battery model, and battery life management | |
| Zhang et al. | SOC estimation of lithium-ion battery pack considering balancing current | |
| US10027134B2 (en) | Active equalization method and system of lithium iron phosphate battery pack | |
| Pei et al. | Online peak power prediction based on a parameter and state estimator for lithium-ion batteries in electric vehicles | |
| CN113777510B (en) | A method and device for estimating state of charge of a lithium battery | |
| CN110525269B (en) | SOC battery pack balance control method | |
| CN113419187A (en) | Lithium ion battery health estimation method | |
| CN110850298B (en) | Data-driven SOH estimation method and system for lithium batteries | |
| CN111142036A (en) | Lithium ion battery online rapid capacity estimation method based on capacity increment analysis | |
| Sun et al. | Study of parameters identification method of Li-ion battery model for EV power profile based on transient characteristics data | |
| Gao et al. | State‐of‐charge estimation and active cell pack balancing design of lithium battery power system for smart electric vehicle | |
| CN110303945B (en) | A method for self-adaptive optimal balance control of battery pack power | |
| Wang et al. | Battery available power prediction of hybrid electric vehicle based on improved Dynamic Matrix Control algorithms | |
| CN112350400A (en) | Lithium battery pack non-hierarchical active and passive equalization circuit and method | |
| CN112630662A (en) | Power battery SOH estimation method based on data driving and multi-parameter fusion | |
| CN112305423A (en) | Lithium ion power battery state of charge estimation method, device, medium and equipment | |
| CN113945852B (en) | Method for evaluating inconsistency of storage battery pack | |
| Sepasi et al. | SOC estimation for aged lithium-ion batteries using model adaptive extended Kalman filter | |
| CN115186579A (en) | Lithium battery health state estimation method based on capacity increment variable quantity curve | |
| CN113807039A (en) | Power state prediction method of series battery system | |
| CN112706655B (en) | Energy balance control method and system for new energy vehicles based on intelligent Internet of Things | |
| CN114325449A (en) | Lithium ion battery SOH prediction method | |
| CN113300016B (en) | Battery charging and discharging control method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20210427 Assignee: Henan Yuzhuo Information Technology Co.,Ltd. Assignor: HUANGHUAI University Contract record no.: X2023980039540 Denomination of invention: Energy balance control method and system for new energy vehicles based on intelligent Internet of Things Granted publication date: 20210907 License type: Common License Record date: 20230815 |
|
| EE01 | Entry into force of recordation of patent licensing contract |